Open-Sourced Reinforcement Learning Environments for Surgical Robotics

• DOI: 10.48550/arxiv.1903.02090
• Publication Date: 2019-01-01
• Authors: Florian Richter; Ryan K. Orosco; Michael C. Yip
• Keywords: FOS: Computer and information sciences; Computer Science - Robotics; 0209 industrial biotechnology; 02 engineering and technology; Robotics (cs.RO)

Reinforcement Learning (RL) is a machine learning framework for artificially intelligent systems to solve variety of complex problems. Recent years has seen surge successes solving challenging games and smaller domain problems, including simple though non-specific robotic manipulation grasping tasks. Rapid in RL have come part due the strong collaborative effort by community work on common, open-sourced environment simulators such as OpenAI's Gym that allow expedited development valid comparisons between different, state-of-art strategies. In this paper, we aim start bridge surgical robotics communities presenting first reinforcement environments robots, called dVRL[3]{dVRL available at https://github.com/ucsdarclab/dVRL}. Through proposed environments, which are functionally equivalent Gym, show it easy prototype implement algorithms problems introduce autonomous precision accuracy assisting, collaborative, or repetitive tasks during surgery. Learned policies furthermore successfully transferable real robot. Finally, combining dVRL with over 40+ international network da Vinci Surgical Research Kits active use academic institutions, see enabling broad fully leverage newest strategies learning, scientists no knowledge test develop new can real-world, high-impact challenges